ＩｏＴ機器向けの軽量化暗号実装技術とユーザ生体継続認証への応用

物联网设备的轻量级密码实现技术及在用户生物识别连续认证中的应用

基本信息

批准号：
19J15225
负责人：
ZHOU LU
金额：
$ 1.22万
依托单位：
The University of Aizu
依托单位国家：
日本
项目类别：
Grant-in-Aid for JSPS Fellows
财政年份：
2019
资助国家：
日本
起止时间：
2019-04-25 至 2021-03-31
项目状态：
已结题

来源：
https://kaken.nii.ac.jp/en/grant/KAKENHI-PROJECT-19J15225/
关键词：
secure authentication support vector machine fuzzy rough sets theory IoT Security Authentication in IoT

项目摘要

In this fiscal year, we combined vector machine algorithms with fuzzy rough sets theory in our secure user biometrics authentication, which is more secure than traditional FaceID and TouchID. Traditional core vector machine (CVM) and support vector machine (SVM) have some limitations when used for data classification, while the addition of fuzzy rough sets theory can dynamically adjust the degree of the membership function, optimizing the weight distribution of each feature, and further improving the classification accuracy.We developed a new algorithm combined with SVM and fuzzy rough sets are used to train and identify malicious domain generated by domain generation algorithms combined SVM with fuzzy rough sets, using online and incremental algorithms to automatically identify and classify non-existent domains as benign or malicious. Experiments show that the algorithm can indeed achieve a high classification accuracy, reaching more than 99%.We also develop a new algorithm combined with CVM and fuzzy rough sets used to train and identify users who login and authenticate through biometric and behavioral characteristics. Our application makes the medical cloud bring more convenience to share medical data within the same hospital or between different hospitals and more secure to unauthorized access. We obtain biological and behavioral characteristics from doctors' own gestures for training and classifying, to ensure that only authorized doctors can access patient data.

本财年，我们将向量机算法与模糊粗糙集理论相结合，进行安全的用户生物识别认证，比传统的 FaceID 和 TouchID 更加安全。传统的核心向量机（CVM）和支持向量机（SVM）在用于数据分类时存在一定的局限性，而模糊粗糙集理论的加入可以动态调整隶属函数的程度，优化各个特征的权重分布，进一步提高分类精度。我们开发了一种结合SVM和模糊粗糙集的新算法，用于训练和识别由SVM和模糊粗糙集相结合的域生成算法生成的恶意域，利用在线和增量算法对非恶意域进行自动识别和分类。现有域为良性或恶意的。实验表明，该算法确实可以达到很高的分类准确率，达到99%以上。我们还开发了一种结合CVM和模糊粗糙集的新算法，用于通过生物特征和行为特征来训练和识别登录和认证的用户。我们的应用使医疗云为同一医院内或不同医院之间共享医疗数据带来更多便利，并且更安全地防止未经授权的访问。我们从医生自身的手势中获取生物和行为特征进行训练和分类，以确保只有授权的医生才能访问患者数据。